Upload
lamcong
View
216
Download
0
Embed Size (px)
Citation preview
Smartwatches & Biometrics for Personal Health Monitoring
Josh Piemontesi, Braden Siempelkamp, and Gordon Minaker
Disclaimers
• We will be discussing exercise-related topics in this presentation. See your doctor first to see if you are cleared to exercise. We are not doctors.
• The nature of the presentation will involve discussing consumer health products. We will try our best to approach this in an objective and unbiased way.
• We are not experts in all fields.
Objectives
• Overview of Smartwatches & Biometrics
• Discuss Fitness monitoring
• Preventative health & wellness
• Specialized monitoring
• Potential Pitfalls
• Potential Best Uses
What is a Smartwatch?
• Smartwatch - Device that is worn that monitors and tracks biometric parameters (Macridis, 2018)
• Biometrics - relating to or involving the application of statistical analysis to biological data
• For our presentation, we will be referring to all devices worn on the wrist as being either a “smartwatch” or “wearable”
• Some examples of wearables:• Fitbit Charge 2• Garmin Vivoactive 3• Jawbone Up3• Apple Watch Series 3• Samsung Gear S3
Smartwatch Components
• Optical sensor• Heart rate
• Contact sensor• Skin temperature
• Altimeter• Altitude gain or loss
• GPS• Biking/running route and distance
• Accelerometer• Step count• worn/not worn state• Overall physical activity levels• posture
• Gyroscope• posture
Purpose of Smartwatches
• Support individuals in becoming more active and lesssedentary (little or no physical activity; e.g., sitting)• Decreased risk of chronic physical and mental health
conditions (Owen, 2010; Warburton, 2006)• Cardiovascular Disease, stroke, hypertension, depression, anxiety,
cancers, diabetes, etc.
• Recommended 150 mins of Moderate to Vigorous Physical Activity (MVPA) per week (Warburton, 2007)
Taken from: https://www.canada.ca/en/public-health/corporate/publications/chief-public-health-officer-reports-state-public-health-canada/2016-health-status-canadians/page-13-what-influencing-health-physical-activity.html
Uses of Smartwatches in Personal Health Monitoring
• Number of steps
• Weekly MVPA mins
• Sedentary time
• Distance walked or ran
• Heart rate
• Flights of stairs
• Calories burned
• Sleep patterns
• Skin temperature
Smartwatch Use
• Danova, 2013:• Projected that 373 million units sold in 2020
• Gartner, 2015:• Projected that 500 million units sold in 2020
Gartner. 2015. “Wearable Technology beyond Smartwatches.” Available at: https://www.gartner.com/smarterwithgartner/wearable-
technology-beyond-smartwatches-3/ (last accessed March 11, 2018)
Danova, T. (2013), “Why the smart watch market is poised to explode as it draws millions of consumers into wearable computing”,
Business Insider, pp. 1-18, available at: http://www.businessinsider.com/global-smartwatch-sales-set-to-explode-2013-9 (accessed
March 11, 2018)
Local Smartwatch Use
• Study in Alberta (Macridis, 2018)• 18 years or older• 1215 participants• Comparison of Normal vs. Overweight and between regions
(Metro Edmonton + Calgary vs. Rest of province)• Leisure-time physical activity (LTPA) questionnaire • Sedentary Behaviour Questionnaire • Questions related to Physical Activity Tracking Devices
Local Smartwatch Use
RESULTS
• Significantly associated with Physical Activity Tracking Devices:
• Being female
• <60 years of age
• Having post-secondary education
• Meeting PA guidelines
• Being overweight/obese
What does this mean?
• People in Canada are interested! • There is a consumer bias. • There are barriers in using wearables.
• 37% <1 month• 35% 1-6 months• 27% >6 months
What do the people want?!
• Many barriers, but...• Perceived usefulness is the most significant variable for
consumers to adopt and accept wearable technology
Quantified Self
https://www.youtube.com/watch?v=8wqC6ad1V_Q
Getting a “STEP” ahead
- Smartwatches have ubiquitous function as a pedometer- Counts the number of steps
- Step counting is one of the most accurate features of smartwatches
Richardson et al., 2008, Annal. Fam. Med.
Smartwatches as Pedometers
How does it work?
- Changes in acceleration in various axial planes
- Tracks pattern of movement of swinging arm; used as a proxy for steps- Complex algorithm
Smartwatches as Pedometers
- Most smartwatches have preset goal of 10 000 steps- What does this mean?
Smartwatches as Pedometers
- The effect of walking 10 000 steps (or any number) varies greatly- Dependent on duration, intensity, & distance
- Pedometer use has been shown to: - Increase physical activity
- Be associated with moderate weight loss (0.7-1.85 kg)- Over 4 wks to 1 yr period
- Have slight increases in cardiovascular performance
Richardson et al., 2008, Annal. Fam. Med.
Smartwatches as Pedometers
- There is more to a healthy lifestyle than 10 000 steps but step goals can help facilitate healthy lifestyle changes
Normal Heart Metrics
Heart Rate (HR)
- Beats per minute
Rhythm
- Should be regular (aka sinus rhythm)
Amplitude
- Size or strength of the pulse- Not currently monitored by smartwatches
Monitoring Heart Rate (HR)
- Resting HR: 60-100 bpm
- Maximum HR estimate:- 220 - age (men)- 226 - age (women)
- Changes throughout lifespan
- Elevated resting HR associated with increased risk of cardiovascular disease and mortality
Amer et al., 2014, Am. J. Cardiology
Taking Your Own Heart Rate
1. Find your radial pulse2. Feel for the radial artery3. Count the number of beats in a set
time interval (20 sec)4. Multiply by 3 to get beats per
minute
Monitoring Heart Rate
- Smartwatch monitoring - real-time tracking- Use an optical sensor to monitor blood flow
- Accuracy varies depending on device- Most within 80-90% of actual HR- Chest strap > wrist monitoring
El-Amrawy et al., 2015, Healthcare Informatics Research
Monitoring Heart Rate
Fitness - track heart rate for aerobic vs anaerobic exercise
- Aerobic intensity- Light: 30-39% HR max- Moderate: 40-59% HR max- High: 60-85% HR max
- Useful for training and exercise- Jogging, walking, cycling, swimming, etc.
Monitoring Heart Rate
Health - general monitoring for healthy patients or patients with known heart conditions (cardiopathies)
- Stay within physician-recommended, safe HR thresholds during rehabilitation and exercise
Monitoring Heart Rhythms
- There is more to the heart monitoring than just the rate
- Heart rhythm, or the regularity of the heartbeat, is an important variable
- Rhythm commonly assessed using an electrocardiogram (ECG)
What is an ECG?
- An ECG (or EKG), or an electrocardiogram, records the electrical activity of the heart
- Gives information on the structure and function of the heart
P
QRS
TAtria
Ventricles
Monitoring Heart Rhythms
- FDA approval for ECG (EKG) sensor for apple watch (Nov. 2017)- Doesn’t mean it is a good device
- Pairs with smartphone application
- Can monitor abnormal heart rhythms (arrhythmias) including atrial fibrillation
What is Atrial Fibrillation?
- An irregular heartbeat (arrhythmia) due to rapid, irregular atrial contractions - “quivering”
American Heart Association; Canadian Heart & Stroke Foundation
Normal Atrial Fibrillation
Importance of Atrial Fibrillation
- Blood stasis in heart can lead to pooling and clot formation
- 1/3 strokes after 60 are caused by A-fib
- A-fib patients have 3-5 times greater risk of ischemic stroke
American Heart Association; Heart & Stroke Foundation
Smartwatch ECG Monitoring
- Back to the smartwatch ECG…. how well does it work?
William et al., 2017, Eur. Heart. J.
Smartwatch ECG Monitoring
- A-fib detection:- Good sensitivity (96%) and specificity (94%) for detecting A-
fib compared to physician-interpreted ECGs- n=52; 225 paired ECGs; mean age 68
- This means that the smartwatch ECG was able to accurately detect both the presence and absence of atrial fibrillation
William et al., 2017, Eur. Heart. J.
Smartwatch ECG Monitoring
- Accurate detection compared to physician interpreted ECGs- 96% of ECGs were interpretable by physicians- 85% symptom-rhythm correlation (Newham et al., 2017)
- Pts. recorded when they were having symptoms
- Patients found device easy to use & lessened anxiety about A-fib
Newham et al., 2017, JAFIB
Smartwatch ECG Monitoring
Drawbacks
- Device is expensive (~$200) and requires a subscription for full range of features ($99/yr)- Limited accessibility to user
- Current studies limited by small sample size and potential bias
- Cannot replace current diagnostic standards but may complement as a tool to monitor patients with known A-fib or arrhythmias
Apple Smartwatch Study
- Partnered with Stanford Medicine
- Via “Apple Heart Study” app, - Data is collected on irregular rhythms
- Users notified if A-fib detected
- Free consult with study doctor and ECG for monitoring
Smartwatch ECG Monitoring
Implications
- Exciting technology and may have future implications in monitoring of patients with heart problems
- May allow for rapid treatment of patients experiencing A-fib
Sleep Tracking
• Sleep disorders seriously affect quality of life
• Insomnia associated with impairments in daytime
functioning (Baglioni, 2013)
• Polysomnography is used for sleep monitoring (Kuo, 2017)
• Limitations of polysomnography (Kuo, 2017; Schute-Rodin,
2008)
• Must be in lab setting
• Electrodes and wiring
• Expensive equipment
Sleep Tracking
• Wearables (actigraphy) in sleep research and sleep medicine
• Measures movement, activity, and sleep
• Several algorithms proposed; each device has its own
measurement characteristics (Tilmanne, 2009; Jean-Louise,
2001; Sazonov, 2004)
• Validation and reliability studies required
• Studies have established this in infants, children, and
adults (Sadeh, 2011; Domingues, 2014; Ustinov, 2013)
• Biggest issue is detecting wakefulness within sleep periods
(Kuo, 2017)
What about validity and reliability for individuals with sleep-related disorders?
Sleep Tracking
• Kuo et al. (2017):
• actigraphy recorder with low sampling rate
• 81 participants
• Polysomnography compared to proposed actigraphy
recorder
• Objective sleep measurements:• Sleep efficiency• Total sleep time• Sleep onset time• Wake after sleep onset
Sleep Tracking
• 22 subjects used to develop sleep-wake algorithm• 43 subjects with ‘good’ sleep + 16 subjects with ‘poor’ sleep
• <85% sleep efficiency• Demonstrated reliability using limited information to estimate
wake-sleep stages during overnight recordings
Circles = “good” sleep
Crosses = “Poor” sleep
Line = Represents linear relationship
Closer to line = more accurate
measure (approximates
polysomnography)
Sleep Tracking
• Limitations for this study:• Unable to classify light sleep, deep sleep, and rapid eye
movement• Involuntary body movements could be misclassified as a
wake stage• Subjects with a long quiet-wake interval can be
misclassified as sleep
• Potential use for:• In-home screening• Evaluation of sleep• Measurement and tracking with treatment
Weight Loss
• Tracks calories burned (energy expenditure) via accelerometer algorithms
• Can use in conjunction with phone calorie tracking applications (e.g., MyFitnessPal) and body weight scale
• Calories in, calories out
Weight Loss
• Stanford University School of Medicine researchers (May 2017) looked at energy expenditure of 7 devices (Apple Watch, Fitbit Surge, Samsung Gear)• 31 women and 29 men• Compared energy expenditure to gold standard
• Most accurate device off by 27%• Least accurate off by 93%
Monitoring Blood Pressure
• Omron BP Monitor• FDA Approved
• Accuracy: +/- 3 mmHg or 2%
• Hypertension is the “silent killer”• Diagnosis is most accurate when
home blood pressure monitoring is used.
• Trends in BP can be useful• Physician review of data
Diabetes
• Monitoring Glucose levels• Less invasive than current options?• Continuous glucose monitoring vs. on-demand• Newer models anticipated to be release soon; 10-day wear.
Falls in the Elderly
• Many apps.• Often used in conjunction with sensors
In smartphone.
• Can notify family.• Reduces fears of falling.• Accuracy not well known.
Epilepsy
• “Neutun” Software
• Medication Reminders
• Journal Log
• Notifies family during emergencies.
Accuracy and Validity
- Devices are marketed under premise that they will improve health & fitness
- But majority of manufacturers provide no empirical evidence- Reliant on external institutions for proper evaluation
Accuracy and Validity
Steps > Distance > Physical Activity > Energy Expenditure > Sleep
Evenson et al., 2015, J. Behavioural Nutrition and Physical Activity
Accuracy and Validity
- In medicine, we only perform a test if we believe it will be useful
- With wearables, patients will be able to monitor themselves constantly
- This will increase the amount of false positive and false negative results
Reliability
- An effective medical tool must be reliable
- A number of pitfalls plague smartwatches in regards to reliability
Reliability
Limited Battery Power
- Reported issue in many studies- Battery life variable; 2 - 7 days with use
- Speculated that power consumption will only increase
Reeder et al., 2016, J. Biomedical Informatics
Reliability
Watch Placement
- HR not collected when sensor is not in contact with skin
- Variations in HR even with skin contact
- Detection difficulties with hand-limited activities (biking) or tasks with high levels of wrist action (washing hands)
Kamdar et al., 2016, Pac. Symp. Biocomput.
Reliability
Data Quality
- Data must be interpretable to be useful- Gaps in data due to:
- Charging- Short breaks- Missed data
Kamdar et al., 2016, Pac. Symp. Biocomput.
Reliability
Interdevice reliability
How consistent are two of the same devices?
- High interdevice reliability for steps, distance, and energy expenditure
- High variation between devices when sleeping- Evaluated when wearing two devices
Evenson et al., 2015, J. Behavioural Nutrition and Physical Activity
Privacy & Security
• Data Ownership• Insurance Companies• Personal Safety - especially in children• Employee Monitoring• Local vs. Cloud data storage
Privacy & Security
• Data Ownership• Insurance Companies• Personal Safety - especially in children• Employee Monitoring• Local vs. Cloud data storage
Smartwatch Summary
Today we have discussed the many applications of smartwatches
Future applications are exciting but caution should be used when using smartwatches for medical purposes
- Further research and validation is required
Currently, smartwatches cannot replace standard healthcare but may be complementary